Guide: The click was cheap.
The job is still open.
Impressions, clicks, starts, applications: each number answers a different question, and none of them alone tells you whether the job gets filled. Here's what you measure where, what the numbers leave out, and where your view ends. Sources included.
The short answer
One number alone
proves nothing.
The click was cheap, but the application never came. Or it came, and nobody was a fit. In the end both show up as one number, and that number tells you neither. So measuring social recruiting means this: for every stage from click to hire, you know where you measure it and where your view ends.
In the ads, you see impressions, clicks, application starts and submitted applications. Screening, interviews and hires live in the ATS. Cost per application comes from the ads, while cost per hire needs data from the ATS. Both help, but they're not the same thing.
It often goes wrong when you compare numbers that don't count the same thing: not every click is a click, and not every application is an application. We don't quote industry benchmarks, because we found none we could back up. Instead, you get the questions to ask of every number.
How costs add up is covered in the cost guide, and how applications get into the ATS in the ATS integration guide. Terms like CPA, application start or attribution are explained in the glossary.
The funnel
You can't see
every stage yourself.
| Stage | Where you measure it | What happens along the way |
|---|---|---|
| Impression | In the ad platform. | An impression is an ad being served, not a person. How many people actually saw the ad is a different number. |
| Click | In the ad platform. | Platforms count several types of clicks, such as all clicks and link clicks. Only compare like with like. |
| Career page view | In the site's analytics or via pixel. | This is where the first number gets lost: load abandonment, ad blockers and missing consent. That's why clicks and page views rarely match. |
| Application start | As an event on the career page or in the form, for Lead Ads in the platform. | The event has to fire cleanly. If the form sits on a different domain at the ATS, you can only reach it by getting changes made there. |
| Submitted application | For Lead Ads in the platform, otherwise via thank-you page, event or feedback from the ATS. | What's counted is the application, not the fit. |
| Screening, interview, hire | In the ATS or at the client. | The ads only see this if someone reports the status back. |
Metrics
The number is right.
The conclusion often isn't.
| Metric | What it shows | What it leaves out |
|---|---|---|
| Click-through rate (CTR) | How often someone clicked an ad that was served. | Whether it turns into an application. Sensational ads can drive up click rate and drag down quality. |
| Cost per 1,000 impressions (CPM) | What delivery costs. | Whether it paid off. It swings with competition, season and audience size. |
| Cost per click (CPC) | What a click costs. | What happens next. A cheap click is worth little if nobody fills in the form. |
| Cost per application start | What a started form costs. | Who dropped out. A start isn't an application yet. |
| Cost per application | Ad spend divided by submitted applications. | Whether the application is a fit. Unqualified applications push the number down and still don't help. |
| Application rate | The share of visitors or clicks that apply. | The denominator. Clicks, page views and starts as the base give you three different rates. Form length has a strong effect, too. |
| Abandonment rate | The share of people who start but don't submit. | The why. Look at mobile and desktop separately. |
Time to hire, cost per hire
Two numbers,
one misunderstanding.
Time to Hire in the DGFP 2025 study is the time from the first applications or first contact to offer acceptance. Time to Fill runs from the decision to fill the role to the start date. Other sources set start and end differently. The terms aren't standardized, so every number needs its definition.
Cost per hire is the average cost of one hire. Usually you add up internal and external recruiting costs and divide by the number of hires in the period. Ad spend alone is not cost per hire. We couldn't view the wording of the SHRM standard, so the formula comes from secondary sources.
With few hires, cost per hire swings wildly. One month with a single hire says little. Values across quarters hold up better.
And in practice? In the DGFP 2025 study (736 people in HR roles, not representative), almost 74% track Time to Hire, but only 39% consider it especially relevant.
Cost per application and cost per hire are not the same thing. The first comes from the advertising, the second needs hires from the ATS.
Measurement
So nothing counts twice
and nothing gets lost.
Pixel and Conversions API. The pixel reports events from the browser, the Conversions API from the server. If the same application arrives via both routes, the platform should count it only once. On Meta that works when event_id and event name match, and deduplication applies for 48 hours (Meta for Developers).
UTM parameters. If you use them, you need utm_source, utm_medium and utm_campaign, or your report shows "(not set)". Values are case-sensitive, and Google recommends lowercase. The campaign name in the link should match the one in the platform (Google Analytics Help).
Feedback from the ATS. Meta lets you send events from the server after the fact. For offline events, the timestamp can be at most 7 days old, and older ones make the whole request fail (Meta for Developers). A hire made weeks later can't simply be reported this way. Whether and how this works in recruiting is something to clarify with your vendor and your applicant tracking system (ATS).
One definition of "application start". Decide up front what triggers it: the opened form, the first filled-in field or the submission of the first page. Otherwise you compare numbers that count different things.
Consent
Why platform and careers page
rarely count the same thing.
General information, not legal advice. For specific questions, ask your data protection officer.
No consent, no pixel. § 25 TDDDG requires it for storing and reading information on the end device. The German Data Protection Conference (DSK) stresses that this is technology-neutral and not limited to cookies. Whether reach measurement without consent is permitted isn't answered in general by the guidance, because the term is undefined (DSK, guidance on digital services (Orientierungshilfe Digitale Dienste), version 1.2, November 2024).
What we didn't find: a statement in the guidance on Meta Pixel and Conversions API specifically. So we don't give a legal assessment. Your data protection officer does.
What it means for your numbers: People who don't consent don't show up in browser tracking. Visits, starts and applications are missing there, yet the platform still counts their clicks. We couldn't back up how big the gap is, so we don't name a consent rate.
Who collects the consent? On the careers page, whoever runs the page. Clarify that before you put numbers from different sources side by side.
Attribution
The ad works.
It doesn't always get the credit.
The attribution window. According to Meta's help, the default is "7 days after click or 1 day after view", and 28-day windows are no longer supported for active campaigns.
Applications take their time. People who see an ad often apply later and come back straight to the careers page. The ad only gets credit for that if it falls inside the window. So its effect tends to be underestimated. We haven't found a study on the size of this effect in recruiting.
On Meta, employment is a special ad category. Age, gender, postal code and lookalikes are restricted, details are in Meta targeting for job ads. With fewer targeting levers left, differences between campaigns more likely come from creative, radius and careers page. That's the editorial team's assessment, not a measurement.
Apple doesn't make it easier. Since iOS 14.5, measurement data from Apple devices is restricted. How Meta handles this today is in its current help. We don't name any numbers.
Reports
Everyone wants
different numbers.
| Who reads | What they want to know | What belongs in the report |
|---|---|---|
| Recruiting team, every week | Which job is running, where does it stall? | Impressions, clicks, starts and applications per job and channel, cost per start and per application, drop-off rate, budget spent. |
| Hiring department, every month | Are the right people applying for my job? | Applications per job, review rate, invited candidates, open jobs and time to hire per job, with definition. |
| Management, every quarter | What does it cost to fill a job, and is it working? | Total ad budget, hires through social media, cost per hire with a disclosed definition, time to fill. |
For every metric, state the definition, period, data freshness and source, meaning ad account or ATS. Flag values based on small numbers.
At HiCandidate™
What you see in the dashboard,
and what stays with you.
The dashboard shows impressions, clicks, application starts and applications, plus the costs from click to application, for every channel and every campaign. Application starts and submitted applications are counted separately.
Filter and export: by job, location and channel. Data is usually refreshed several times a day. Reports come as Excel or PDF, and the data is available via API.
Where our view ends: What happens in the ATS after an application is submitted, we can't see. That data sits with you. Suitability, interviews and hires therefore aren't in the dashboard. You get the cost per application from us, cost per hire only with your ATS data.
Not described on this page, and so not claimed by us here: how an application start is tracked technically, which attribution window applies and how gaps from missing consent are handled. Ask in the demo.
More on the Dashboard and reporting page.
Sources
Where we looked it up,
and where we didn't.
As of November 2026. We checked statements with numbers or quotes against the named source, as far as it was accessible. We label vendor sources as such.
- Meta for Developers: Conversions API (platform documentation): how the Conversions API works
- Meta for Developers: Handling Duplicate Pixel and Conversions API Events (platform documentation): removing duplicates, event_id, 48 hours
- Meta for Developers: Offline Events (platform documentation): sending window of 7 days
- Meta for Developers: Special Ad Categories (platform documentation): special ad category for employment
- Google Analytics Help (help page): UTM parameters, required fields, spelling
- German Data Protection Conference (DSK) (guidance on digital services, version 1.2, November 2024): consent under Section 25 TDDDG (German Telecommunications Digital Services Data Protection Act)
- DGFP, HTWK Leipzig, Wollmilchsau (study "Recruiting-Strukturen, ein Benchmark", 2025, 736 respondents, not representative): terms time to hire, time to fill, cost per hire, prevalence
- Meta Business Help: attribution models (vendor information): default attribution window
What we don't know:
No industry benchmarks. For CPC, CPM, cost per application, click-through rate and application rate in social recruiting, we found no study with a traceable method. That's why we give no reference values.
No consent rates and no size of the measurement gap. There's no reliable source for these either.
No legal advice. We don't give a legal assessment of the pixel, Conversions API or consent.
Frequently asked questions
Questions about
measuring in social recruiting.
Which metrics matter in social recruiting?
The ones that map the path from click to hire: impressions and clicks as a first step, then application starts, submitted applications and their cost. After that, in the ATS, screening, interviews and hires. None of these numbers stands on its own. What matters is that each has a fixed definition and the same period.
What's the difference between cost per application and cost per hire?
Cost per application is the ad spend divided by the submitted applications. Cost per hire is the total cost of a hire, internal costs included, divided by the number of hires. One comes out of advertising, the other exists only with data from the ATS.
Where does measurement end in social recruiting?
Mostly at the submitted application. Screening, rejection, interview and hire live in the ATS. A vendor only sees them if the customer reports them back. That includes HiCandidate™.
Why do clicks in the ad platform and visits to the career page differ?
Because they don't count the same thing. A click isn't a loaded page view yet, and without consent, tracking scripts don't fire on the careers page, while the platform still counts the click. We haven't been able to show how big the gap is.
Do I need consent for measurement?
For storing and reading information on the end device, § 25 TDDDG usually requires consent, technology-neutral and not only for cookies. The German Data Protection Conference (DSK) says so in its guidance. We found nothing there on Meta Pixel and Conversions API specifically. Clarify that with your data protection officer.
Which numbers belong in a recruiting report?
That depends on the recipient. The recruiting team needs starts, applications and costs per job and channel every week. The business unit asks about applications, invited candidates and time to fill per job. Management needs total budget, hires and cost per hire with a disclosed definition every quarter.
What does HiCandidate™ measure and what doesn't it?
The dashboard shows impressions, clicks, application starts and applications, plus the costs up to the application, for every channel and every campaign. What happens in the ATS after an application is submitted, HiCandidate™ doesn't see. Suitability, interviews and hires therefore aren't in the dashboard.
Demo
Numbers you understand?
We'll show you the dashboard.
In 30 minutes, we'll show you the dashboard with example data online and answer your questions about measurement.
Book a demo →